CN1113619A - Method and device for processing a signal - Google Patents

Abstract

A method and device for processing a signal are described, wherein an estimate of a multipath-induced contribution to a demodulated navigation signal is calculated and subtracted from said demodulated navigation signal to obtain an estimated line of sight contribution to said demodulated navigation signal, and a propagation time tau 0 is calculated from the thus calculated line of sight contribution to said demodulated navigation signal, such that a very accurate propagation time tau 0 of said navigation signal can be calculated.

Description

Method and device for processing a signal

The present invention relates to a kind of method and apparatus of processing signals, for example be used for communication system or satellite navigation system, the method and apparatus of the processing signals that especially the present invention relates to can be eliminated any error that causes owing to multipath or make it at least to reduce.In a specific embodiment, the present invention relates to navigation system.

From the principle, comprise by transmitter by the communication between the transmitted from transmitter to receiver and to launch by the emission of information signal by the code modulated carrier signal of mode of amplitude modulation or frequency modulation, and by the reception of the signal of receiver received code.In receiver,, the signal that receives is decoded in order to obtain information wherein.X time and the signal between the time of reception are propagated institute's elapsed time and are depended on the distance between transmitter and receiver and the propagation velocity of carrier signal, and this also is well-known.

In fact, it is not only so directly to propagate by sighting distance that carrier signal arrives receiver, but also passes through such as building, mountain, cloud, the one or more reflection in these objective objects of atmosphere or the like and reach receiver.Because the propagation path through reflection always is longer than sighting distance, therefore always long through the required time of line-of-sight propagation than signal through the reflections propagate time, that is to say that the signal of reflection always has certain time-delay with respect to the signal of direct propagation.Receiver can not be differentiated direct signal and reflected signal, the direct signal that receives by receiver with combine signal if not distortion with processed by the reflected signal of receiver, yet this said binding signal comprises the error with respect to the signal of emission significantly.Problem that Here it is known " multipath ".Multipath also can be caused by diffraction.

The communication of mentioned kind is generally used for a variety of application, and it generally wishes to be used to reduce because the error that multipath produces in most application.Yet reducing error is navigation to its very important a kind of application.For example be satellite navigation, the present invention is described with reference to navigation system, does not limit to such use but should conclusively show the present invention.

In contextual should using, navigation is according to the principle of determining the distance between object and reference position with signal of communication.A base station, satellite for example, the signal that its emission is received by the receiver relevant with said object.By the propagation time of measuring-signal, can calculate the distance of this object to base station.Similarly, the distance of this object to the second and base station as much as possible also can be calculated.Under static base station situation, the position of base station is constant, and can add receiver as constant value.Under the situation of astatic base station, for example be satellite.This signal packet contains the information of relevant base station position.Therefore, the position of base station is " known " for receiver, so can calculate the position accurately of object.

The navigation system called after GPS(global positioning system of the known type of above-mentioned mistake) and rowland-C(remote navigation system-C) these systems for example are used for deepwater ship and by the aircraft utilization near the airport.In order more at large to describe these systems, will be with reference to following publication:

" satellite digital communication " author J.J.Spilker delivered at the prentice Hall of New Jersey in 1977;

" worldwide navigation: GPS users' guidebook ", author N.Ackroyd and R.Lorimer, nineteen ninety is published on the Lloyd ' s of london publishing society in London;

" radio navigation system ", author B.Forsell delivered on prentice Hall (ISBN 0-13-75 058-6) in 1991;

Can be clear that from above-mentioned application accuracy is very important, and since the error that causes of multipath should eliminate or will reduce as much as possible at least.

Therefore, main purpose of the present invention is to eliminate the error that multipath causes.

The method and apparatus that produces according to the present invention is used to handle the signal of reception, thereby realized these some purposes of the present invention, said method and apparatus has ability and the base value of definite visual signal and the ability of multi-path signal base value that has multipath that detect.So method and apparatus of the present invention is removed the base value of multi-path signal from the signal that receives, thereby has kept a undisturbed visual signal distance.Therefore, use this non-visual signal distance of being disturbed, can determine from the signal source to the receiver signal, to propagate used correct time.

By discussion, describe and explain the present invention with reference to accompanying drawing in more detail to preferred embodiment.

Fig. 1 shows and uses sight of the present invention;

Fig. 2 shows the shape of distortionless navigation signal;

Fig. 3 shows related operation;

Fig. 4 shows the Utopian shape of correlation function and the usual way of handling navigation signal;

Fig. 5 shows the combination at three correlation function peaks of multi-path signal and the shortcoming of usual way;

Fig. 6 shows the step that the best approach of the present invention is taked;

Fig. 7 a illustrates the Utopian shape of correlation function, and the method for processing navigation signal of the present invention;

Fig. 7 b illustrates the shape of auxiliary function;

Fig. 8 shows effect of the present invention;

Fig. 9 shows the example according to a device of the present invention;

Figure 10 shows the shape of the correlation function in the reality.

Fig. 1 shows an afloat ship 1 and 4 the satellite 2 of transmitting.Generally represent in order to the satellite equipment that produces and transmit with 3.The structure of satellite equipment 3 does not constitute a part of the present invention, and the knowledge of this structure is dispensable for understanding the present invention.Therefore satellite 2 and satellite equipment 3 are not done further discussion.As long as saying satellite 2 can be that presently used gps satellite is just much of that.

For current discussion, will ignore the data that signal 4 comprises the information of carrying, and only consider that navigation signal 5 is modulated on the carrier wave of signal 4.Fig. 2 shows the shape by the navigation signal 5 of equipment 3 emissions.In Fig. 2, the signal strength signal intensity I that ordinate is represented, unit is arbitrarily, abscissa express time t.Navigation signal 5 shows with the individual pulse form, and for example in rowland-C system employed (curve A), perhaps navigation signal 5 is with (curve B) of the impulse form appearance of a sequence, for example is used for gps system.Navigation signal 5 shown in Fig. 2 is periodically launched, and cycle P is the 1ms(millisecond in said system).It is illustrated in the t of time shaft _OInstantaneous each signal period begins emission, and this stipulates constantly very exactly, and it is by the clock control of system.

Curve A among Fig. 3 has been represented the GPS navigation signal 5 under bigger time scale, only shows the first of this signal here.GPS navigation signal 5 has constituted the Pseudo-Random Noise Code that comprises predetermined code figure place.For sign indicating number position is distinguished mutually with data bit, claim usually sign indicating number be " breach " (chip), each breach can have two possible values, promptly+1 or-1.Said predetermined number is 1023 in gps system.Breach separately will be referred to as C(1 below), C(2) ... C(1023).Each satellite 2 all has by one group of C(1), C(2) ... C(1023) its condition code of defined.

Receiver 10 on ship 1 has a code generator at least, and it can make it to produce the corresponding specifically reference signal of the condition code of satellite 2.Usually, receiver 10 is equipped with a memory, the information stores of the condition code of a plurality of satellites is on this memory, so, it is that he is interested that the operator only need indicate that satellite, and the code generator of receiver 10 will automatically produce the reference signal (adjustment code generator) of the condition code of corresponding interested concrete satellite.

Fig. 3 and Fig. 4 show detection and will be referred to as delay time T as mentioned above ₀The principle in propagation time of signal 4.The curve A of Fig. 3 shows as the navigation signal 5 by satellite 2 emission, just C(1) beginning and the time started t of time cycle ₀Coincide.Therefore, the curve A of Fig. 3 also can be taken the embodiment of the reference signal that is produced by the code generator of receiver 10 as.Curve B is illustrated in navigation signal 5 ' (not having noise and multipath), just C(1 that the input 11 of receiver 10 receives) beginning relative time t ₀Postponed time τ ₀It should be noted that carrier wave and data do not illustrate in Fig. 3.

In order to determine delay time T ₀, the unshowned control device relevant with receiver 10 provides a local reference signal 6, and it is identical with reference signal 5, but has been offset a definite shift time τ.This local reference signal 6 is represented by the curve C of Fig. 3.

In order to determine whether that this (known) shift time τ equals this (the unknown) time delay τ ₀, the navigation signal 5 that said control device receives local reference signal 6 and input 11 at receiver 10 ' compare.More particularly, local reference signal 6 carried out by control device 6 and the navigation signal 5 that receives ' multiplying, ground of local reference signal 6 pointwises just and the navigation signal 5 that receives ' multiply each other.Therefore, this consequential signal represented by curve D in Fig. 3, and is referred to as product signal 7, and this consequential signal is by average, just product signal 7 in enough big time integrates so that the mean value M(τ of product signal 7 is provided).

Product signal 7 can only get+1 or-1 value, this depend on signal 6 and 5 ' value whether mutually the same or different.Therefore, the mean value M(τ of product signal 7) will be+value of 1 and-1, because the shape of product signal 7 depends on shift time τ, M(τ) value accurately depend on the selective value of shift time τ.Because navigation signal 5 is pseudo noise codes, for the major part value of shift time τ, mean value M(τ) will almost be zero, if shift time τ is near delay time T ₀, mean value M(τ) and will be different from zero significantly.In ideal conditions, if shift time equals delay time T ₀, said mean value M(τ) and will equal 1.Curve E among Fig. 3 represents the average value of a function M(τ as shift time τ) this characteristic, this function is referred to as correlation function 8.

The shape accurately of known in advance correlation function 8, and depend on the concrete code of the satellite in talking about.Curve E in Fig. 3, shown in shape be desirable triangle.Leg-of-mutton wide W is known constant value, for practical meaning, equals the duration of the breach of twice; Therefore, correlation function 8 complete peaked coordinate (τ by it ₀), M(τ ₀) determine the τ here ₀It is the associated arguments of determining to be relevant to the position of the correlation function 8 that the time cycle begins.

It shows that correlation function 8 in the reality is not the desirable triangle shown in the curve E of Fig. 3, but more or less is crooked, as shown in figure 10, and it depends on the characteristic of receiver.

In addition, it also shows for example in rowland-C system, have only a navigation signal pulse rather than a pseudo noise code with reservation shape to be launched in each time cycle; As discussed earlier, navigation signal pulse itself can substitute correlation function 8.

Hereinafter, being combined in Fig. 4 explanation of representing correlation function 8 on the bigger ratio is used for detecting and definite propagation time τ ₀The general method in common receiver, at first, for first numerical value t of shift time τ ₁Produce local reference signal 6, and can calculate the mean value M(t of corresponding product signal 7 ₁).Secondly, second to shift time is worth t ₁+ △ t produces local reference signal 6, and calculates the mean value M(t of corresponding product signal 7 ₁+ △ t).Here the sampling interval of being represented by △ t is a constant value.

From dividing other M(t ₁) and M(t ₁+ △ t) numerical value can calculate error signal ₁=M(t ₁)-M(t ₁+ △ t).This error signal is in order to the shift time of control local reference signal 6, so ought be subsequently to two value t ₂And t ₂+ △ t calculates mean value M(τ) time, error signal ₂=M(t ₂)-M(t ₂+ △ t) less than ε ₁The process of this known delay lock loop (DLL) is carried out continuously, so any concrete instantaneous ε ≈ 0 in practice, and supposes it is symmetrical based on the peak of correlation function 8, can calculate propagation time τ according to following formula ₀,

τ ₀＝t ₁+1/2△t。

Usually the critical defect of this method is the propagation time τ that calculates ₀Enough inaccurate, can give explanation in conjunction with Fig. 5, the shape of illustrated correlation function 23 is that correlation function 24 combines with correlation function 25 and 26, correlation function 24 be as the navigation signal 5 by line-of-sight reception ', correlation function 25 and 26 be as by reflect the navigation signal 5 that receives '. Reflection peak 25 and 26 propagation time τ ₁And t ₂Propagation time τ greater than line of sight peak 24 ₀, shown they divide other maximum A ₁And A ₂Maximum A less than line of sight peak 24 ₀ Reflection peak 25 and 26 width W ₁And W ₂Be substantially equal to the width W of line of sight peak 24 ₀

In usual way, the processing of correlation function 23 is used as and is not had multipath, and just to be based on it be the hypothesis of distortionless correlation function and handling to correlation function 23.Usual way for example can cause I(t _x) and I(t _x+ △ t _x) satisfy the stopping criterion of error signal and propagation time τ _xCan be by τ _x=t _X+ 1/2 △ t _xCalculate, this will clearly find out τ _xThe propagation time τ that is not equal to line of sight peak 24 ₀

In fact, because peak 23 is asymmetric when multipath is arranged, with the time τ of usual method calculating _xIt even not the accurate estimation of the maximum value position at peak 23.For sampling peak 23 on the higher position at peak, improving usual way by this respect that reduces △ t.This for example can cause I(t _y) and I(t _y+ △ t _y) satisfy the stopping criterion of error signal, according to τ _y=t _y+ 1/2 △ t _yCalculate propagation time τ _y, τ _yNear the estimation of the maximum value position at peak 23, but still be not equal to τ _O

In fact, though usual way is improved, so that the peak at accurate Calculation peak 23, draw correct τ as shown in Figure 5 in the situation _OIt also is because its inherence former thereby impossible can be clear that in Fig. 5 the maximum of the relative line of sight peak 24 of the maximum of resultant peak 23 has been offset that value is used for usual way.

What shown the performance of these peaks in Fig. 5 is desirable form, just sharp-pointed summit, however as suggested in preceding, summit is crooked in practice, this will cause and be relevant to τ ₀The deviation of evaluation in propagation time increase further.

The present invention is based under the situation that has multipath navigation signal 5 ' and the mathematical understanding of the true form of correlation function 23.Therefore, eliminated the influence of multipath according to an aspect of the present invention, perhaps make it to reduce to minimum at least, this be by to the navigation signal 5 that receives ' the measurement of complete shape of correlation function 23, the influence of the pair correlation function that causes of estimation multipath and the effect of sighting distance pair correlation function and calculate propagation time τ by the effect of the sighting distance pair correlation function that calculates ₀

In discussion subsequently, will explain P(t) be a function of expression navigation signal 5, the reference signal in the memory of its corresponding stored in receiver 10.Here suppose reflected signal (multipath) Mac to the navigation signal 5 that receives ' correlation function 23 influential; And each said reflective function Mac can be by function x _i(t)=a _iP(t-τ _i) cos(ω t+ θ _i) expression, wherein

a _iIt is the amplitude that is designated as the signal of i down;

τ _iIt is the delay time that is designated as the signal of i down;

θ _iIt is the phase place that is designated as the signal of i down;

ω is the angular frequency of the carrier signal of navigation signal.

Yet, the navigation signal 5 of the reality that is received by receiver 10 ' can write is being arranged under the propagation condition of multipath:

$\mathrm{r(t)=}\underset{\mathrm{i=0}}{\overset{\mathrm{Mac}}{\mathrm{\Σ}}}{x}_i\mathrm{(t)+n(t)=}\underset{\mathrm{i=0}}{\overset{\mathrm{Mac}}{\mathrm{\Σ}}}\mathrm{ai·P(t-\tau i)·cos(\omega \; t+\theta i)+n(t)}$ （1）

The subscript i=0 here represents line-of-sight signal, and subscript i=1,2 ... Mac represents multi-path signal.In addition, n(t) expression noise component(s).

In fact signal x _i(t) also taken advantage of by data-signal, it is by making it this signal and estimated data's signal multiplication remove.Very clear those skilled in the art, this is because data-signal must be removed from common intrinsic delay lock loop systems.For reduced representation, data-signal has omitted in formula (1).

Can observe a _i, τ _i, θ _iBe known variable, these variablees are represented signal with subscript i, and its time to time change.According to maximum possibility estimation theory, known parameters a _i, τ _i, and θ _iBest estimation

With

Be those numerical value, those value minimizings L[S(t)], minimizing is defined as:

$\mathrm{L\; ［s(t)］=}{\∫}_O^T{\mathrm{［r(t)-s(t)］}}^2\mathrm{dt}$ （2）

The S(t here) is defined as

It shows in practice, and the signal is here taken a sample, top integral operation can by one determine sample value that the time interval is all with substitute.

With the criterion of above equivalence be L(S(t)) all partial derivatives be zero.Can see that this criterion will satisfy:

Here

X} is defined as the τ for X(), the numerical value of τ has the maximum of its integral body $R_x\mathrm{(\tau )=}\frac{2}{T}{\∫}_O^T\mathrm{r(t)·P(t-\tau )·exp(-j\omega \; t)dt}$

Be homophase (real part) and quadrature (imaginary part) down-conversion correlation function, the combination correlation function 23 of the example that discussed corresponding front; And R(τ) is homophase (real part) and quadrature (imaginary part) benchmark correlation function, the orthogonal function 8 that corresponding case discuss is in front crossed, defined correlation function will appear in the actual receiver, and to its normalization, so that maximum at this peak is 1 and phase place is 0, as shown in Figure 3.Preferably when the measurement of benchmark correlation function, there are not noise and multipath according to the present invention, and use a signal simulator, and this shape is stored in as the benchmark correlation function in the memory of receiver, for example with the form of form or with the form of the function of representing said shape.

We will can be clearly seen that in case equation (4) solves, purpose of the present invention has just reached, and this is because the delay time T of line-of-sight signal ₀Can think by

Accurately calculate and.Yet equation (4) has been represented optimal parameter for the occurrence of Mac

Between iterative relation.Can directly solve an equation in theory (3) though use Iterative Matrix to calculate, this approaching expensive computing time uses the computer system of costliness extremely fast and large-scale to make real-time implementation only become possibility fully.

In actual process of the present invention, with reference to the iterative computation of Fig. 6 description to equation (4).

In the 1st step 101, according to formula R ₀(τ)=R _X(0651), the 1st peak correlation function R ₀(τ) approach and equal R _X(τ).

In the 2nd step 102, calculate conduct at R with a kind of mode that will describe later on ₀The parameter of the optimum value of the maximum peak (τ)

With

In the 3rd step 103, according to formula

From R _XDeduct said parameters calculated (τ)

With

Base value and export the 2nd peak function R ₁(τ).

In the 4th step 104, parameters calculated

With

As R ₁The optimum value of the maximum peak (τ)

In the 5th step 105, according to formula

From R _XDeduct said parameters calculated (τ) With

Base value to draw first peak correlation function R ₀New being similar to (τ).

In the 6th step 106, calculate the parameter that makes new advances

With As the new approximate R that in the 5th step, calculates _OPeak-peak optimal values (τ).

The 3rd, 4,5 and 6 steps repeat to be satisfied until suitable stopping criterion 201.For an example of suitable stopping criterion 201, said step can repeat predetermined times, for example is 10.Yet, also may repeat said step up to parameter

With According to predetermined convergence criterion convergence, this also be those of ordinary skills institute clearly.The result's that proof draws convergence criterion detects the delay of whether estimating between two iterative steps in succession

And/or

Variation less than 0.1 nanosecond (ns).

In this, be the navigation signal 5 that receives ' the provide base value of estimation for two signals

With

, as

Be inphase/orthogonal down-conversion correlation function R _XThe base value R of the estimation that (τ) provides ₀(τ) and R ₁(τ), as

If the base value to the estimation of the 3rd signal is calculated, according to formula

, as in the 7th step 107, from R _XDeduct parameter (τ) With

The base value of calculating to export the 3rd peak correlation function R ₂(τ).

In the 8th step 108, parameters calculated

As R ₂The optimum value of maximum peak (τ)

In the 9th step 109, the new parameter that calculates

As

In the optimum value of maximum peak.

In the 10th step 110, use is calculated in the 9th step

The parameter τ that calculating makes new advances ₀, a ₀, θ ₀As

In the optimum value of maximum peak.

In the 11st step 111, use is calculated in the 9th step

With in the 10th step, calculate

Calculate new parameter

As

In the optimum value of maximum peak.

Repeat the 9th, 10,11 step and satisfied until stopping criterion 202, it can be equivalent to stopping criterion 201.

With simulated mode, the new value at the base value at the 4th peak and three initial peaks can be calculated iteratively.The new value at the base value of the 5th peak value and four peaks at first can be calculated iteratively then, or the like, this also be this field those of ordinary skill institute clearly.

Describe by different way, the base value of all signals does not just calculate simultaneously from beginning, but only there is the base value of a signal at first to calculate, the base value of the 2nd signal adds thereon then, and the base value of two signals is best, add the 3rd signal then, and the base value of all three signals is best, or the like, can carry out this process continuously and calculate until the base value of all signal Mac+1.Yet it shows that the number Mac of the actual multi-path signal that occurs does not know in advance in practice, even needs not to be constant.Therefore, after each cycle of said process after the base value of M+1 signal is calculated, the M here is considered to the number of multipath base value, if M is less than Mac, the multipath error can not be eliminated fully, otherwise, if the noise error that M greater than Mac, has increased will be introduced into.So, according to the present invention, allow M=0 successively, 1,2 in order to address this problem ... M _Max, preferably solve an equation (4) as previously explained, in case suitable stopping criterion is satisfied, just stop this iterative process, perhaps work as M=M _MaxNo matter which at first occurs.In order to prevent that computing time is oversize, M _MaxIt is the predetermined fixed numeric values that is used for a upper limit of conduct of iterative process.Suitable numerical example such as M _Max=3 or M _Max=4.

This shows that in the process of foregoing description, the peak of first calculating preferably has subscript 0, otherwise previous subscript 0 is in order to the indication line of sight peak, and this peak needn't the corresponding peak that at first calculates.Below, the peak of sighting distance will be indicated once more with subscript 0.It shows that after said process was finished, one group of parameter of corresponding line of sight peak was easy to identify, and this is because it is to have One group of parameter of minimum value.

Increase in case detect error, said iterative processing just can stop, as what represent by the suitable stopping criterion of formula (5):

SRP（M）≤SRR（M-1）

Then, select for the resulting parameter value of M-1 as optimum value.

In formula (5), according to

Be defined as signal-residual ratio.After having estimated all parameters, the V(M here) is defined as residual deviation according to formula (7).

The D here is a delay aperture, delay interval just, and this comprises the peak that sighting distance relevant peaks and multipath disturb at interval.In fact, the suitable length of D is the perdurabgility of the breach of twice, and the position of suitable D is such, so that the sighting distance relevant peaks of estimation is located substantially on the center of D, perhaps is positioned within first half of D.This be since only length of delay during greater than the length of delay of sighting distance relevant peaks the multipath base value can estimate.

This has also shown the correlation function R of inphase/orthogonal down-conversion in practice _X(τ) sampled, thus the integral operation of formula (7) can with and computing replace.

As another stopping criterion, can use formula (8) to replace formula (5)

SRR（M）≥δ

The δ here is a suitable threshold, for example can select it to equal concrete signal-noise ratio SNR, and SNR exactly can cause receptible error

Another stopping criterion can be represented by formula (9)

The max(lresl here) be defined as peaked absolute value at interval D residue functions res, res is defined as:

Implementing simply when of the present invention, verified under most of actual conditions, M _MaxThe fixing number that selection equals 1 can produce enough good result.In this case, for two peaks the strongest, promptly the approximate parameter value calculation of line of sight peak and the strongest reflection peak has gone out.Almost be that remainder error is only actual important under the identical situation only in the intensity of two or more reflection peaks.

Below calculating parameter will be described

As correlation function R _iThe optimum process of the optimum value of the maximum peak (τ), described as reference example as step 102.

At first be to determine that of compound sampling of said correlation function is for [Re(R _i(τ))] ²+ [Im(R _i(τ))] ²Maximum is provided.Pairing delay will be called τ _MaxIn other words:

τ _max＝ (max)/(τ) {[Re（R _i（τ））] ²+[Im（R _i（τ））] ²} （11）

From a said sampling and according to formula (12) evaluation phase :

In case known this phase place, in order to obtain intrinsic modulation signal according to formula (13), this phase place will be in order to do the phase place rotation on input signal.

Therefore the phase modulated signal that should obtain is in order to estimated delays and amplitude.

Can in all sorts of ways and estimate the delay at peak

Two kinds of basic diverse ways will be discussed below.

At first can use all samplings at peak, and the interior difference of calculating these samplings, this is in order to obtain the more or less continuous performance of the shape at peak.Then, use the known analytical method of Newton-Raphson method to calculate peaked position.If Nyquist criterion is satisfied in the sampling interval, this method provides the delay estimation that is relevant to noise bias the best.

Secondly, it also can only use the sampling of a small amount of at peak.This method is based on the shape that will know the peak in advance, and preferably according to the present invention, this is because it is more faster than previously described method.It is 2 so little that the number of said sampling resembles, and discusses below in conjunction with Fig. 7.

Fig. 7 A shows a correlation function, and it can be compared with correlation function among Fig. 48.The delay τ at peak ₀Expression, the sampling A and the B of this function represents to adopt two different t time of delay ₁And t ₂, sampling interval △ τ is with t ₁And t ₂Separate, so that t ₂=t ₁+ △ τ, the sizes related R ' of measurement _i(t ₁) and R ' _i(t ₂) use R respectively ₁And R ₂Expression.

Below, exactly be in t ₁And t ₂Middle delay τ _OExpression, so that t ₁=τ _C-1/2 △ τ and t ₂=τ _C+ 1/2 △ τ.

In addition, t _CDistance τ _xExpression, so that t ₁=τ ₀-τ _x-1/2 △ τ and t ₂=τ ₀-τ _x+ 1/2 △ τ.Because τ ₀With △ τ be constant, the relevant numerical value R ' that clearly measures _i(τ _C-1/2 △ τ) and R ' _i(τ _C+ 1/2 △ τ) can be regarded as τ _xFunction.

With the correlation values R ' that measures _i(τ _C-1/2 △ τ) and R ' _i(τ _C+ 1/2 △ τ) calculate form parameter f, it defines with formula (14).

${\mathrm{f(\tau }}_x\mathrm{)=}\frac{{\mathrm{R(\tau }}_o{\mathrm{-\tau }}_x+\frac{1}{2}{\mathrm{\Delta \tau )-R(\tau }}_o{\mathrm{-\tau }}_x-\frac{1}{2}\mathrm{\Delta \tau )}}{{\mathrm{｜R(\tau }}_o{\mathrm{-\tau }}_x+\frac{1}{2}{\mathrm{\Delta \tau )｜+｜R(\tau }}_o{\mathrm{-\tau }}_x-\frac{1}{2}\mathrm{\Delta \tau )｜}}$ (14)

Therefore, f also is τ _xFunction, in Fig. 7 B with curve representation f(τ _X), vertical axis unit arbitrarily.Clearly, be the function that descends continuously at f in peaked certain scope, it shows at each τ _xValue and the value of each f between be 1: 1 relation.The shape of function f (τ) depends on the shape of correlation function, the size of the characteristic curve of receiver and △ τ.As previously mentioned,, also can calculated in advance go out function f (τ), calculate inverse function f owing to can set up the shape of correlation function in advance for each receiver ^-1And this inverse function is stored in the memory of receiver, for example as the form of a form or polynomial repressentation.

Between operational stage, according to formula (14), two sampling value R ' at peak _i(τ _C-1/2 △ τ) and R ' _i(τ _C+ 1/2 △ τ) is used for calculating f.Use the inverse function f of storage ^-1, then calculate τ _x

So, τ ₀Be estimated as

T preferably ₁And t ₁+ △ τ selects as far as possible near the peak.Especially on the opposite at peak, this is because it will produce the most accurate result.

As another, can be according to formula (15) definition f.

$\mathrm{f=}\frac{{\mathrm{R(\tau }}_c+\frac{1}{2}\mathrm{\Delta \tau )}}{{\mathrm{R(\tau }}_c-\frac{1}{2}\mathrm{\Delta \tau )}}$

In another embodiment, use 4 sampling values, the f here can according to formula (15a) and 15(b) definition:

${\mathrm{f(\tau }}_x\mathrm{)=}\frac{{\mathrm{R(\tau }}_o{\mathrm{-\tau }}_x+\frac{3}{2}{\mathrm{\Delta \tau )-3R(\tau }}_o{\mathrm{-\tau }}_x-\frac{1}{2}{\mathrm{\Delta \tau )+3R(\tau }}_o{\mathrm{-\tau }}_x+\frac{1}{2}{\mathrm{\Delta \tau )-R(\tau }}_0{\mathrm{-\tau }}_x+\frac{3}{2}\mathrm{\Delta \tau )}}{N}$ (15A)

or

${\mathrm{f(\tau }}_x\mathrm{)=}\frac{{\mathrm{R(\tau }}_o{\mathrm{-\tau }}_x{\mathrm{-2\Delta \tau )-2R(\tau }}_o{\mathrm{-\tau }}_x{\mathrm{-\Delta \tau )+2R(\tau }}_o{\mathrm{-\tau }}_x{\mathrm{+\Delta \tau )-R(\tau }}_0{\mathrm{-\tau }}_x\mathrm{+2\Delta \tau )}}{N}\mathrm{(15b)}$

The N here is the normalization parameter, equal some measured values absolute value and.For example,

N=|R(τ ₀) | perhaps N=|R(τ ₀-1/2 △ τ) |+| R(τ _O+ 1/2 △ τ) |.

In theory because calculus of differences plays similar deconvolution, the width that reduces pulse with folded pulse is become more separate, the sort of third order difference method provides the insensitive advantage of multipath.On the other hand, the inexactness in receiver.Especially in local reference signal, influence is more seriously realized result's accuracy.

As another formula (13), it can be according to for example formula (13a) or (13b) uses incoherent restituted signal:

${\mathrm{R\prime i(\tau )=［Re(Ri(\tau ))］}}^2\mathrm{+［Im(Ri(\tau ))］}{}^2$ （13a）

$\mathrm{R\prime i(\tau )=}\sqrt{\mathrm{［Re(Ri(\tau ))］}{}^2\mathrm{［Im(Ri(\tau ))］}{}^2}\mathrm{(13b\; ）}$

The amplitude of maximum peak in signal

The quickest way of estimation be to select of the estimation of maximum sampling value simply as amplitude.The error that causes can realize acceptable so little, for example by selecting △ τ arbitrarily little.Also can use delay as far as possible

To top described calculating with the control sampler, so that the sampling at peak exactly occurs in the top of the calculating at (very approaching) peak.

For the estimation amplitude, also can use the technology of similar above-mentioned relevant delay as much as possible.Normalization peak shape function g can be defined as follows:

g(τ)= (R(τ))/(R(τ _o)) （15）

Use the amplitude sampling x of a measurement and the delay of estimation

Make above-mentioned method and calculate, can estimate amplitude according to following formula.

As another, can use the second dervative at peak, according to:

g(τ _x)= (R(τ _o-c _x-Δτ)-2R(τ _o-τ _x)+R(τ _o-τ _x+Δτ))/(N) （17）

And carry out according to the similar approach that resembles the relevant formula of describing the front (15).

Favourable influence of the present invention will be described by the discussion to example in conjunction with Fig. 8.

The gps antenna of standard is placed on the optional position of base (or foot) of big building, receives the navigation signal of GPS at some time durations.This signal is handled in two receivers simultaneously, and first receiver is the GPS receiver of standard, the principle work of being discussed according to reference Fig. 4, however another receiver carries out work according to principle of the present invention.M _MaxSelection equals 1, when between two iterative steps in succession Variation less than 0.1ns, perhaps after 10 iterative steps, no matter that takes place earlier, this iteration will be stopped.The number of (compound) sampling is 20.Distance between sampling, 1/2 of a breach duration.As 10 initial samplings, 1,/10 10 samplings that are used as subsequently of a breach duration.Formula (14) is used for f.Be 1 second the time of integration.

Fig. 8 shows the delay time T of the line of sight peak that two receivers calculate ₀Fluctuating δ _T, according to formula δ _L=C δ _τBe converted into range measurement δ _L, and relative time t depicts.When time t=0, the value of calculating is as fiducial value, and is made as zero (0); The value that t in Fig. 8＞0 o'clock is described is calculated, and is the value that deducts from fiducial value.

Solid line is illustrated in the fluctuating that takes place in the calculating of GPS receiver of standard.Because multipath apart be 60 meters between the extremum of the fluctuating that this calculated distance is measured, and the deviation of the calculating range measurement of standard is greater than 10 meters.

The fluctuating that second receiver that produces according to principle of the present invention occurs is illustrated by the broken lines, though, may be because noise, the calculated distance measurement remains fluctuating, extremum is only at a distance of 5 meters, and the standard deviation of error is reduced to 1.1 meters, and promptly improvement factor is 10.

Below, will describe the exemplary embodiments of processing unit 30 of the present invention in conjunction with Fig. 9 in detail.

Receiver 10 comprises an input (antenna) 11 in order to receive navigation signal 4 as mentioned above.The signal 4 that receives is sent to the input 31 of processing unit 30, and input 31 is coupled to the input 35 of multiplier 34.It shows that if desired, before sending the 1st input 35 of multiplier 34 to, the signal 4 of reception can be exaggerated and/or down-convert to a predetermined intermediate frequency, and then to its digitlization.This is known to the those skilled in the art; For the cause of simplifying, amplifying device, down-conversion device and digitalizer do not show in Fig. 9.

It is the signal generator 32 of the signal of ω that processing unit 30 comprises in order to produce frequency, and an output of generator 32 is coupled to the 2nd input 36 of multiplier 34.The signal 4 that receives in multiplier 34 is by demodulation, and the navigation signal 5 of demodulated received ' be provided on the output 37 of multiplier 34.

It has shown processed signal has been processed into composite signal, for this reason generator 32 has two outputs in fact at least, a signal that provides can be called Cos(ω t), relative the 1st signal of another signal that provides has the phase difference of pi/2, and this signal can be called Sin(ω t).This Cos(ω t) signal is used for handling reality (homophase) portion of the signal of reception, opposite Sin(ω t) be used for handling void (quadrature) portion of the signal of reception.On two kinds of processing procedure principles is identical.Therefore,, in Fig. 9, only show only output 33 of generator 32, and will be referred to as exp(j ω t after being combined in of said output signal for the cause of simplifying), this be this field those of ordinary skill institute clearly.

Processing unit 30 has many multipliers of N 40 ₁, 40 ₂, 40 ₃Be called multiplier 40k below the 40N, by 1 to N interval, k is an integer, and the N correspondence here is to the number of times of navigation signal 5 ' sampling of receiving.Each multiplier 40k has the 1st input 41k, the 2nd input 42k and an output 43k.Each the 1st input 41k of multiplier 40k is coupled on the output 37 of multiplier 34, with navigation signal 5 ' r(t) the exp(j ω t of the reception of receiving demodulation).

Processing unit 30 has many integrator 45k of N.Each integrator arrangement 45k has an input 46k who is coupled to the output 43k of relevant multiplier 40k, with one of the input 51k that is coupled to digital signal processor 50 output 47k, it can be a computer this be those skilled in the art institute clearly.Integrator arrangement 45k arranges to sue for peace in order to the sampled signal that will receive at its input 46k during predetermined period T, and with the output 47k that this summation operation result delivers to it, opens new summation operation again from zero then.

Digital signal processor 50 has the clock input 52 of the output 56 of being coupled to precision clock device 54.In addition, digital signal processor 50 has control output 55, and it is in order to controlled driving local reference signal generator 60, and the output 63 of local reference signal of providing 6 is provided generator 60.Local reference signal generator 60 has the benchmark input 61 of being coupled to memory 62, has stored the information of the feature code of relevant navigation signal 5 in memory, and this erect image is foregoing.Though in Fig. 9, do not illustrate, memory 62 comprises a plurality of feature codes of relevant different satellites, and local reference signal generator 60 can selective reception the information of a said code only, that just selectively be tuned to the navigation signal of predetermined satellite, this also is that this field those of ordinary skill is known and known.

Processing unit 30 comprises many deferred mounts of N-1 70 ₁, 70 ₂, 70 ₃70 _N-1, being referred to as deferred mount 70n afterwards, n is an integer in 1 to N-1 interval.Each deferred mount 70 _nAn input 71 is arranged _nWith an output 72 _n, this device is arranged in order to the output 72 at it _nProvide and postpone to have surpassed D time of delay _{τ n}The input 71 at it _nReceived signal duplicate model.In one embodiment, all deferred mount 70 _nIdentical D time of delay is provided _τThese deferred mount serial connections, so that deferred mount 70 _nInput 71 _nBe coupled to previous deferred mount 70 _N-1Output 72 _N-1First deferred mount 70 ₁Input 71 ₁Be coupled to the output 63 of local reference signal generator 60, it also is coupled to the 1st multiplier 40 ₁The 2nd the input 42 ₁Each deferred mount 70 _nOutput 72 _nBe coupled to relevant multiplier 40 _N+1The 2nd the input 42 _N+1

The inclined to one side journey of digital signal processor 50 quilts is to drive local reference signal generator 60 with synchronous exactly by the 54 official hour cycles of clock apparatus.Yet just as explained later, digital signal processor 50 can be offset first breach C(1 of local reference signal 6) the beginning t in beginning relative time cycle ₀With certain offset THS _bSo,, each multiplier 40 _kIn its 2nd input 42 _kReceive local reference signal 6, the beginning t in this signal relative time cycle _OSurpass shift time τ _k

According to formula $\mathrm{ck=\tau b+}\underset{\mathrm{n=1}}{\overset{\mathrm{k-1}}{\mathrm{\Σ}}}{\mathrm{D\tau }}_n$

The local reference signal 6 of this skew can be expressed as:

pk（t）＝p（t-τ _K）

In its output 43 _k, each multiplier 40k provides product signal 7, and it can be expressed as:

r（t）·exp（jωt）·p（t-τ _K）

Therefore, each integrator 45k provides a signal at its output 47k, and it is expressed as:

$\frac{1}{T}{\∫}_{\mathrm{t=o}}^T\mathrm{r(t)·exp(j\omega \; t)·P(t-\tau k)}$

And it is at inphase/orthogonal down-conversion correlation function R _XThe selected sampling in τ (τ)=τ k place.

Said sampling is refreshed behind each integration period T, and in its a plurality of inputs 51 _kRead by digital signal processor 50.Digital signal processor 50 group of equations (4) of programming are preferably used the iterative step of describing in conjunction with Fig. 6, have realized the input 51 at it in this step _kRead said sampling.Under the situation of gps system, selected integration period T preferably is no more than 10 milliseconds, and this is in order to obtain suitable data estimation.Yet this does not mean that digital signal processor 50 calculates whole equation group (4) and definitely must be 10 milliseconds at every turn, is that ionospheric variation has play a part important here.For group of equations (4) per second once or even usually be each less time.

More particularly, in the situation of a gps system, with usual way estimation carrier phase can be per 10 milliseconds once.From the phase place of this estimation, can determine the data of local generator and the deviation of frequency.Therefore this frequency departure of determining is in order to the frequency of control-signals generator and code generator, so that their frequency error remains zero substantially.The data of the calculating that occurs with the phase jitter form are by multiply by coherent sampling the data bit of estimation, so that remove these data from coherent sampling, the coherent sampling of being taken advantage of after this can be averaged in 1 second or longer time.In order to obtain the accurate estimation of phase place and delay, per second once or even usually be that each less time is enough to can group of equations (4).

According to the phase place of calculating and the estimation of delay, digital signal processor 50 control generators 32 and local reference signal generator 60, generator 60 preferably comprises numerically controlled oscillator.Generator 32 Be Controlled more particularly, consequently Be approximately zero.By revising offset THS _b, local reference signal generator 60 is so that the method that the peak remains in the scope of delay aperture D is controlled.In addition, may be with such method control local reference signal generator 60, promptly the maximum phase place of received signal correlation function is zero.For more information, relate to Spilker, he has described tracking loop and delay lock loop, they are designed in order to guaranteeing that in the sample one top with the line of sight peak of estimation coincides, and two other sample be in said peak symmetry around.

Digital signal processor 50 also has a data output 59, it in order to the estimated value of presenting calculating to the appropriate display device, such as plotter, printer or CRT, and/or give such as the such suitable storage device of the known computer storage in this field.In fact, this is the feed-in computer as a result, and this computer also receives the estimation result who is used for some other transmitter, and this is in order to calculate the position accurately of receiver.It is identical that these calculate the calculating of carrying out with existing receiver.

For perfect cause, it shows two multipliers 34 in practice, first received signal Cos(ω t at first at their the 2nd input 36 places separately), second received signal Sin(ω t), thus their output 37 separately provide respectively homophase navigation signal 5 ' and quadrature navigation signal 5 '.In addition, it also shows device 40; The dual appearance of 45 and 70 circuit, so as to handle respectively homophase navigation signal 5 ' and quadrature navigation signal 5 '.

We can revise the very clear embodiments of the invention of having described those of ordinary skill in the art under the situation of the scope of not leaving scope of the present invention or claim and improve it.For example, be enough to only to handle reality (homophase) portion of received signal in this case, formula (4) is simplified to formula (4A)

In addition, be to successively decrease continuously or increasing function as long as be used for the function of f, just can be with other suitable function as the form parameter f and the g that had described.

Claims (17)

1, in comprising the signal transmission system of transmitter and receiver, the method in order to processing signals in said receiver comprises the following steps:

Received signal;

The signal of demodulate reception;

Obtain the sampling R of signal of the reception of decoding at a plurality of time τ place _X(τ);

The estimation to the base value of demodulated received signal of calculating that multipath introduces, from the signal of demodulate reception, deduct base value that said multipath introduces obtaining the vision base value, and calculate between transmitter and receiver signal and propagate used propagation time τ apart from estimation to the demodulated received signal _o, said propagation time τ _oBe the vision base value distance after calculating said demodulated received signal to be calculated.

2, in comprising the signal transmission system of transmitter and receiver, the method in order to processing signals in said receiver comprises the steps:

Received signal;

The signal of demodulate reception;

Reference signal is provided;

In order to obtain correlation function, the signal that receives is carried out related operation with reference signal;

Obtain the sampling R of correlation function at a plurality of time τ place _X(τ);

Calculate multipath and introduce the estimation of base value pair correlation function;

Deduct the estimation of base value that the multipath of estimation introduces from correlation function with the vision base value distance that obtains pair correlation function; With

Calculating signal between transmitter and receiver is propagated used propagation time τ _O, said propagation time τ _OCalculating of pair correlation function obtains from the vision base value distance of calculating.

3,, it is characterized in that also comprising in the described estimation steps parameter according to the said method of claim 2

With

Calculation procedure, be expressed as follows:

Here $R_x\mathrm{(\tau )=}\frac{2}{T}{\∫}_O^T\mathrm{r(t)·P(t-\tau )·exp(-j\omega \; t)dt}$ In

R(τ) is the benchmark correlation function;

With

Be respectively a _i, τ _iAnd θ _iEstimated value.

4,, it is characterized in that also comprising in the described estimation steps parameter that is expressed as follows according to the said method of claim 2 With Calculation procedure:

Here

R(τ) is the benchmark correlation function;

With

Be respectively a _i, τ _iAnd θ _iEstimated value;

5, according to the said method of claim 3, it is characterized in that the benchmark correlation function here is under the situation that does not have noise and multipath, use signal to simulate that device measures, and the shape of the benchmark correlation function of the measurement here for example is to be stored in the memory of receiver with the form of form or with the form of the function of representing said shape.

6,, it is characterized in that the estimation steps here also comprises the step that execution is following according to the said method of claim 2:

A) calculating parameter

With

As R _XThe optimum value of the maximum peak (τ);

B) according to following formula: from R _XDeduct the parameter τ that is calculated (τ) _O, a _OWith

With

Base value R ₁(τ)

C) calculating parameter τ ₁, a ₁, θ ₁As R ₁The optimum value of the maximum peak (τ);

D) according to following formula from R _XDeduct parameters calculated τ (τ) ₁, a ₁, θ ₁Base value to draw R _ONew approximation (τ),

E) calculate new parameter

As the new approximation R that in step (d), calculates _OThe optimum value of the maximum peak (τ); With

F) repeating step (b), (c), (d), (e), up to satisfying suitable criterion.

7, according to the said method of claim 6, it is characterized in that the step (b) here, (c), (d), (e) repeat predetermined times.

8, according to the said method of claim 6, it is characterized in that the step (b) here, (c), (d), (e) repeat up to

With

At least one parameter is consistent with the convergence criterion of being scheduled to.

9,, it is characterized in that here in order to estimate effect, execution the following step to the 3rd signal according to the said method of claim 6:

G) according to following formula from R _XDeduct parameters calculated (τ)

With

Base value to draw

,

H) calculating parameter

As R ₂The optimum value of the maximum peak (τ);

I) calculate new parameter

As

In the optimum value of maximum peak;

J) resemble calculating in step (i), operation parameter

Calculate new parameter As

In the optimum value of maximum peak;

K) resemble calculating in step (i), operation parameter

Calculate new parameter

As

In the optimum value of maximum peak; With

L) repeat (i), (j), (k) step is satisfied up to stopping criterion.

10, according to the said method of claim 2, it is characterized in that:

According to:

A sampling R from said peak _i(τ _Max) estimation peak phase place

, preferably but be not must or near the top at this peak.

11,, it is characterized in that also comprising calculated in advance function f (τ according to the said method of claim 2 _X) step, the inverse function f here ^-1Calculate and be stored in the memory of receiver.

12,, it is characterized in that the function f (τ (X)) here is defined as according to the said method of claim 11:

${\mathrm{f(\tau }}_x\mathrm{)=}\frac{{\mathrm{R(\tau }}_o{\mathrm{-\tau }}_x+\frac{1}{2}{\mathrm{\Delta \tau )-R(\tau }}_o{\mathrm{-\tau }}_x-\frac{1}{2}\mathrm{\Delta \tau )}}{{\mathrm{｜R(\tau }}_o{\mathrm{-\tau }}_x+\frac{1}{2}{\mathrm{\Delta \tau )｜+｜R(\tau }}_o{\mathrm{-\tau }}_x-\frac{1}{2}\mathrm{\Delta \tau )｜}}$

13,, it is characterized in that the function f here is defined as according to the said method of claim 11:

$\mathrm{f=}\frac{{\mathrm{R(\tau }}_c+\frac{1}{2}\mathrm{\Delta \tau )}}{{\mathrm{R(\tau }}_c-\frac{1}{2}\mathrm{\Delta \tau )}}$

According to the said method of claim 11, it is characterized in that the function f here is defined as

${\mathrm{f(\tau }}_x\mathrm{)=}\frac{{\mathrm{R(\tau }}_o{\mathrm{-\tau }}_x+\frac{3}{2}{\mathrm{\Delta \tau )-3R(\tau }}_o{\mathrm{-\tau }}_x-\frac{1}{2}{\mathrm{\Delta \tau )+3R(\tau }}_o{\mathrm{-\tau }}_x+\frac{1}{2}{\mathrm{\Delta \tau )-R(\tau }}_0{\mathrm{-\tau }}_x+\frac{3}{2}\mathrm{\Delta \tau )}}{N}$

The N here is the normalization parameter.

14,, it is characterized in that the function f here is defined as according to the said method of claim 11:

${\mathrm{f(\tau }}_x\mathrm{)=}\frac{{\mathrm{R(\tau }}_o{\mathrm{-\tau }}_x{\mathrm{-2\Delta \tau )-2R(\tau }}_o{\mathrm{-\tau }}_x{\mathrm{-\Delta \tau )+2R(\tau }}_o{\mathrm{-\tau }}_x{\mathrm{+\Delta \tau )-R(\tau }}_0{\mathrm{-\tau }}_x\mathrm{+2\Delta \tau )}}{N}$

The N here is the normalization parameter.

15, according to the said method of claim 14, it is characterized in that the N here equal some measured values absolute value and, for example

${\mathrm{N=｜R(\tau }}_o\mathrm{)｜orN=｜R(\tau }{}_o-$ 1/2 Δτ)｜+｜R(τ ₀+ 1/2 Δτ)｜

16,, it is characterized in that the sampling R here according to the said method of claim 2 _X(τ) obtain simultaneously.

17, the device of processing signals comprises:

Digital signal processor;

It is ω that generation has frequency, is expressed as exp(-j ω) the generator of signal, this generator is by DSP CONTROL.

First multiplier has the first input end and second input, and the reception navigation signal is coupled in first input, and the output of generator is coupled in second input;

A plurality of N second multipliers, each of second multiplier have the first input end that is coupled to the output of first multiplier;

A plurality of N integrator arrangement, each integrator arrangement are coupled to the input of relevant second multiplier output and the output that is coupled to the digital signal processor input;

The local reference signal generator is coupled to the control output end of digital signal processor, and has an output at least local reference signal is provided;

A plurality of N-1 deferred mounts, each deferred mount have an input and an output, and each output is connected to second input of the second relevant multiplier;

The input of first of said a plurality of deferred mounts here is coupled to first second input of the output of local reference signal generator and said a plurality of second multipliers, and the output of previous deferred mount is coupled in the input of another deferred mount here.

Images